基于联合观测的移动机器人多传感器融合定位方法
Multi-Sensor Fusion Localization for Mobile Robots Based on Joint Observation
张宇祯 1宋原 1郭磊1
作者信息
- 1. 北京邮电大学智能工程与自动化学院，北京100876
- 折叠
摘要
\justifying 针对非结构化环境下移动机器人单一传感器易退化、多源传感器融合精度受限的问题,提出一种基于扩展卡尔曼滤波的融合定位方法。首先,针对传统异步顺序更新忽略同一时刻多传感器观测噪声相关性的不足,构建融合激光里程计和光流传感器的联合观测状态空间模型。其次,针对传感器安装偏置引起的测量扰动,基于刚体运动学推导安装偏置误差补偿模型,并嵌入观测雅可比矩阵。仿真结果表明：所提算法的定位精度最高,在常规工况下其轨迹均方根误差为0.28~cm。在激光雷达测量值受干扰场景下,该算法仍能维持鲁棒的状态估计,轨迹均方根误差为0.90~cm。
Abstract
\justifying In unstructured environments, mobile robots often suffer from single-sensor degradation and limited multi-source fusion accuracy. This paper proposes an Extended Kalman Filter (EKF)-based fusion localization method. First, a joint-observation state-space model integrating LiDAR odometry and an optical-flow sensor is constructed to overcome asynchronous sequential updates that neglect same-time correlation among multi-sensor observation noises. Second, an installation-offset compensation model derived from rigid-body kinematics is embedded in the observation Jacobian to mitigate offset-induced measurement disturbances. ROS/Gazebo simulation results show that the proposed algorithm achieves the highest localization accuracy, with trajectory RMSE of 0.28~cm under normal conditions and 0.90~cm under LiDAR measurement interference while maintaining robust state estimation.关键词
控制理论与控制工程/移动机器人定位/多传感器融合/扩展卡尔曼滤波/联合观测Key words
control theory and control engineering/mobile robot localization/multi-sensor fusion/extended Kalman filter/joint observation引用本文复制引用
张宇祯,宋原,郭磊.基于联合观测的移动机器人多传感器融合定位方法[EB/OL].(2026-03-30)[2026-03-31].http://www.paper.edu.cn/releasepaper/content/202603-290.学科分类
自动化技术、自动化技术设备
评论